Vector-Valued Control Variates
arXiv:2109.08944
Abstract
Control variates are variance reduction tools for Monte Carlo estimators. They can provide significant variance reduction, but usually require a large number of samples, which can be prohibitive when sampling or evaluating the integrand is computationally expensive. Furthermore, there are many scenarios where we need to compute multiple related integrals simultaneously or sequentially, which can further exacerbate computational costs. In this paper, we propose vector-valued control variates, an extension of control variates which can be used to reduce the variance of multiple Monte Carlo estimators jointly. This allows for the transfer of information across integration tasks, and hence reduces the need for a large number of samples. We focus on control variates based on kernel interpolants and our novel construction is obtained through a generalised Stein identity and the development of novel matrix-valued Stein reproducing kernels. We demonstrate our methodology on a range of problems including multifidelity modelling, Bayesian inference for dynamical systems, and model evidence computation through thermodynamic integration.
arXiv commentAccepted for publication at ICML 2023
References in corpus (8)
- Zero-variance principle for Monte Carlo algorithms
- Zero Variance Markov Chain Monte Carlo for Bayesian Estimators
- Stein Variational Gradient Descent With Matrix-Valued Kernels
- Convex Learning of Multiple Tasks and their Structure
- Neural Control Variates for Variance Reduction
- Scalable Control Variates for Monte Carlo Methods via Stochastic Optimization
- The Controlled Thermodynamic Integral for Bayesian Model Comparison
- A Unifying and Canonical Description of Measure-Preserving Diffusions